# Pegasus: A Universal Framework for Scalable Deep Learning Inference on the Dataplane

arXiv: https://arxiv.org/abs/2506.05779 · HTML: https://mlsanity.com/paper/2506.05779/

## Key facts

- Cited by 2 papers in the map (data through 14 Oct 2026).
- Builds on 21 of its 72 references that are in the map.
- First posted on arXiv 6 Jun 2025.

## Builds on (most cited first, 21 of 21)

- [DoReFa-Net: Training Low Bitwidth Convolutional Neural Networks with Low Bitwidth Gradients](https://mlsanity.com/paper/1606.06160/) (2016)
- [Efficient Processing of Deep Neural Networks: A Tutorial and Survey](https://mlsanity.com/paper/1703.09039/) (2017)
- [Quantized Neural Networks: Training Neural Networks with Low Precision Weights and Activations](https://mlsanity.com/paper/1609.07061/) (2016)
- [Neural Additive Models: Interpretable Machine Learning with Neural Nets](https://mlsanity.com/paper/2004.13912/) (2020)
- [Kitsune: An Ensemble of Autoencoders for Online Network Intrusion Detection](https://mlsanity.com/paper/1802.09089/) (2018)
- [A Sensitivity Analysis of (and Practitioners' Guide to) Convolutional Neural Networks for Sentence Classification](https://mlsanity.com/paper/1510.03820/) (2015)
- [NetLLM: Adapting Large Language Models for Networking](https://mlsanity.com/paper/2402.02338/) (2024)
- [Understanding Convolutional Neural Networks for Text Classification](https://mlsanity.com/paper/1809.08037/) (2018)
- [Multiplying Matrices Without Multiplying](https://mlsanity.com/paper/2106.10860/) (2021)
- [New Directions in Automated Traffic Analysis](https://mlsanity.com/paper/2008.02695/) (2020)
- [NetMamba: Efficient Network Traffic Classification via Pre-training Unidirectional Mamba](https://mlsanity.com/paper/2405.11449/) (2024)
- [Programming Protocol-Independent Packet Processors](https://mlsanity.com/paper/1312.1719/) (2013)
- [Neural Packet Classification](https://mlsanity.com/paper/1902.10319/) (2019)
- [Taurus: A Data Plane Architecture for Per-Packet ML](https://mlsanity.com/paper/2002.08987/) (2020)
- [In-network Neural Networks](https://mlsanity.com/paper/1801.05731/) (2018)
- [You Do (Not) Belong Here: Detecting DPI Evasion Attacks with Context Learning](https://mlsanity.com/paper/2011.01514/) (2020)
- [Brain-on-Switch: Towards Advanced Intelligent Network Data Plane via NN-Driven Traffic Analysis at Line-Speed](https://mlsanity.com/paper/2403.11090/) (2024)
- [Yet Another Representation of Binary Decision Trees: A Mathematical Demonstration](https://mlsanity.com/paper/2101.07077/) (2021)
- [ClickINC: In-network Computing as a Service in Heterogeneous Programmable Data-center Networks](https://mlsanity.com/paper/2307.11359/) (2023)
- [NN-Defined Modulator: Reconfigurable and Portable Software Modulator on IoT Gateways](https://mlsanity.com/paper/2403.09861/) (2024)
- [A Deep Learning Perspective on Network Routing](https://mlsanity.com/paper/2303.00735/) (2023)

## Built on by (most cited first, 2 of 2)

- [Versatile yet Efficient Network Traffic Analysis: Offloading Network Foundation Model to SmartNIC](https://mlsanity.com/paper/2508.02001/) (2025)
- [Chimera: Neuro-Symbolic Attention Primitives for Trustworthy Dataplane Intelligence](https://mlsanity.com/paper/2602.12851/) (2026)

## Abstract (arXiv)

The paradigm of Intelligent DataPlane (IDP) embeds deep learning (DL) models on the network dataplane to enable intelligent traffic analysis at line-speed. However, the current use of the match-action table (MAT) abstraction on the dataplane is misaligned with DL inference, leading to several key limitations, including accuracy degradation, limited scale, and lack of generality. This paper proposes Pegasus to address these limitations. Pegasus translates DL operations into three dataplane-oriented primitives to achieve generality: Partition, Map, and SumReduce. Specifically, Partition "divides" high-dimensional features into multiple low-dimensional vectors, making them more suitable for the dataplane; Map "conquers" computations on the low-dimensional vectors in parallel with the technique of fuzzy matching, while SumReduce "combines" the computation results. Additionally, Pegasus employs Primitive Fusion to merge computations, improving scalability. Finally, Pegasus adopts full precision weights with fixed-point activations to improve accuracy. Our implementation on a P4 switch demonstrates that Pegasus can effectively support various types of DL models, including Multi-Layer Perceptron (MLP), Recurrent Neural Network (RNN), Convolutional Neural Network (CNN), and AutoEncoder models on the dataplane. Meanwhile, Pegasus outperforms state-of-the-art approaches with an average accuracy improvement of up to 22.8%, along with up to 248x larger model size and 212x larger input scale.